Re: [h5md-user] Subgroups

[Felix Höfling]

Am 09.09.2013, 12:43 Uhr, schrieb Konrad Hinsen  
<address@hidden>:

> One of the points that are unclear to me in the current specification
> is the definition of subgroups.
>
> Suppose I have a system consisting of a protein in solvent. I want to
> have two subgroups, "protein" and "solvent". But I also have data that
> applies to the whole system, without breaking it down into subgroups.
> And I might have yet another data set that uses different subgroups
> for a specific purpose.
>
> My understanding of the current specification is that particles
> are numbered from 0 to N-1 inside each subgroup. I would have
> atoms number 0 to K-1 for my protein, and 0 to L-1 for my solvent.
> Then how do I know what these atoms correspond to in the complete
> system? That information cannot be deduced from the data for each
> subgroup, so it has to be supplied somewhere else. But where?
>
> Konrad.

Hi Konrad,

It seems that your question on the subgroups has not yet been answered.

So far, the subgroups have mainly be used for output purposes not to  
reconstruct a complete simulation (except for a few cases with manual  
interference where the user has provided the semantics between the  
subgroups).

Actually, the intention of H5MD is that not all information has to be  
stored (e.g., solvent particles may be absent), in this sense the union of  
all subgroups does not yield the complete set of particles. On the other  
hand, the subgroups are not required to be disjoint subsets. For examples,  
subgroups may be selected according to particle type or to a region in  
space, which makes a lot of sense for the observable tree.

I'm not sure whether overlapping subgroups are also useful for the data in  
"particles". (Maybe one example for a spatial selection: a user wants to  
analyse the interfacial region between two liquids, so only a shallow slab  
along the interface is stored instead of all particles.) Do you see a  
problem here w.r.t. your applications?

Automatic reconstruction of a complete simulation from a H5MD file would  
be nice, but there is much more to it than just particle coordinates.  
Then, one would have to enter into specifying particle interactions in a  
generic way which appears impossible to me.

Felix